In our Nurture project, we conducted two simulation-based experiments to study how can we use reinforcement learning algorithms to model user interruptibility and schedule notifications based on user context. Our paper is accepted in UbiTtention workshop 2018. The title of the paper is "Nurture: Notifying Users at the Right Time Using Reinforcement Learning", co-authored by Bo-Jhang Ho, Bharathan Balaji, Mehmet Koseoglu, and Mani Srivastava.
Note: Unfortuantely, as we did not seek permission for redistribution of the data during the informed consent process, due to IRB IRB restrictions we are unable to provide third party access to the dataset. Our apologies!
Please contact the PI Mani Srivastava for the dataset. Once you get the dataset, you have to make a couple of symbolic links so that the codebase can be aware of these datasets. Trace this repository and it will be mentioned in README.md in different folders.
For the current folder, please make the following symbolic link:
ln -s /path/to/figure/folder export_figures